Extend the server-side profiler code to handle reboots as well. Sadly,
this is a bit of a kludge, but I can't see any better solution without
redesigning the client-side profiler interface first.
I handle a reboot by pulling back any existing profiler data right
after a reboot and then re-starting the profiler. This generally works
because most of the profilers are idempotent; either they write into
different files in each run, or they just continuously append logs
to a log file of some kind and so concatenating all their output
is good enough.
In general there's unfortunately no reliable way to signal the
profilers before a reboot occurs, we can only deal with it after the
fact, so we don't have the option of adding some sort of
save-and-restore methods to the client-side interface that could be
used to deal with this.
Right now we simply don't have any real use cases anyway where we want
to run a profiler on top of a reboot test where the profiler requires
special handling after a reboot. Until we actually have an example of
that to work with, trying to implement support for that situation
may just be overdesigning things anyway.
Risk: Medium
Visibility: Server-side profilers will be restarted after a reboot
during a test.
Signed-off-by: John Admanski <jadmanski@google.com>
git-svn-id: http://test.kernel.org/svn/autotest/trunk@2526 592f7852-d20e-0410-864c-8624ca9c26a4
6 files changed